Log of the #fcrepo channel on chat.freenode.net

Using timezone: Eastern Standard Time
* ajs6f joins07:18
* ksclarke leaves07:48
* mhwood joins08:03
* dhlamb joins08:04
* awead joins08:47
* jrgriffiniii joins08:48
* osmandin joins09:01
* awoods joins09:20
* awead leaves09:32
* awead joins09:45
<ajs6f>awoods: When you close 1418 I will examine 1411.10:01
<awoods>ajs6f: What do you mean by "examine"?10:03
ajs6f: 1411 still needs work, no?
<ajs6f>awoods: Peer at it. Close one eye and emit humming sounds. Make a small gesture of disgust and walk away.
awoods: I will see what can be done with what osmandin left behind.
<awoods>ajs6f: perfect. That would be great.10:04
ajs6f: what are your thoughts on a bulk-load capability?
<ajs6f>awoods: what about 1539?
awoods: Loading directly into the JCR?
<awoods>ajs6f: no. Loading an RDF dump that contains lots of F4 resources.10:05
<ajs6f>awoods: Bulk loading to save HTTP cost?10:06
* mhwood leaves
<awoods>ajs6f: yes, as well as to save parsing a bulk RDF document.
<ajs6f>awoods: How are you going to avoid parsing RDF? Are you going to load it with your eyes closed yelling "I can't see you!"?10:07
* whikloj joins
<awoods>ajs6f: see: http://www.theeuropeanlibrary.org/tel4/access/data/opendata/details
<ajs6f>awoods: Those are giant RDF datasets. I don't get it. How do you load those witout parsing RDF?10:08
<awoods>ajs6f: They are examples of single RDF documents that describe multiple resources.
<ajs6f>awoods: Yes, they are.
The Bavarian State LIbrary kicks ass at making triples.10:09
<awoods>ajs6f: There is some user-demand for being able to load such documents into F4 (after changing the URIs to point to F4).
ajs6f: or: http://www.theeuropeanlibrary.org/tel4/access/data/lod10:10
<ajs6f>awoods: Changing URIs is _not_ the repos job. Loading RDF is. This is one good reason to pay attention to our Java API. That's what we use to build this kind of feature.
awoods: Sounds like a pretty simple streaming script. Want me to write it in Scala?10:11
awoods: This is a feature-request about pure RDF? These people don't care about bitstreams?10:12
<awoods>ajs6f: Agreed, it is a fairly simple scripting task. Unfortunately, many F4 users do not have scripting skills.
ajs6f: yes, this is a scenario where F4 holds metadata only.10:13
<ajs6f>awoods: Our long-term plan is to downgrade the software quality and upgrade the users' tech skills until they meet in the middle.
<awoods>ajs6f: new topic: fcr:tombstone
<ajs6f>awoods: Tombstones are dad to me.
dead
<awoods>ajs6f: daddy
<ajs6f>awoods; Or as my boy says, "Papapapappapa!"10:14
<awoods>ajs6f: I need to track down the notes, but fcr:tombstone came out of discussions around handling repository deletes.
ajs6f: the other option on the table was using the status property10:15
<ajs6f>awood: Yes. I get why we did it. I was there. But we know better now. barmintor is with me on this.
<awoods>ajs6f: marking items as deleted.
<ajs6f>awoods: There's a continuum here and one end is a persistent data structure.
<awoods>ajs6f: removing fcr:tombstone is fine... but are you also proposing introducing a status=delete that is recognized by the repo... along with a purge feature?10:16
* awead leaves
<ajs6f>awoods: not as such. I'm proposing that whatever happens should become visible without the construction of a new URI.
<awoods>ajs6f: can you elaborate?10:17
<ajs6f>awoods: There should be no fcr:tombstone. Whatever change in the representation of a resource we want to take place upon its deletion should be visible at the same URI as was the before-deletion representation.10:18
awoods; Personally, I would be happy with an HTTP redirect to whereever. If you wanted to keep the thing you just deleted, you should have versioned it.10:19
<awoods>ajs6f: Have you experimented with deleting a resource that has been versioned?
<ajs6f>awoods: I haven't even run Fedora in months. I try not to use it.10:20
<awoods>ajs6f: My suspicion is that it is gone.
<ajs6f>awoods: Then we can choose not to delete it. We can move it to a garbage can and leave a node in its place with a link to the old thing that appears in HTTP as a redirect. The important point is that it should not be a new URI.10:21
awoods: And if people delete the redirect-thingy, the old resource gets garbage collected out of the can.
<awoods>ajs6f: Having "the representation of a resource we want to take place upon its deletion being visible at the same URI as was the before-deletion representation" sounds very nice. Your fcr:tombstone ticket only addresses half of the issue. The other half is how to support a new approach.10:22
<ajs6f>awoods: ^^^ + ^^^10:23
<awoods>ajs6f: Yes, I do not think we want to remove the fcr:tombstone without something like you are describing being a part of that same effort.
<ajs6f>awoods: Fine. I don't much care about what that is. I think people get too invested in their content already. If you like the idea above, just c+p it to that ticket.10:24
<awoods>ajs6f: It seems like there should be another ticket that is coupled with your new fcr:tombstone DIE ticket.
<ajs6f>awoods: Okay, or c+p it to a new ticket and link 'em.
<awoods>ajs6f: sounds good.
<ajs6f>awoods: What about a ticket for a streaming loader?10:28
<awoods>ajs6f: Do you think it could be generalized?10:29
<ajs6f>awoods: Not sure what you mean...?
<awoods>ajs6f: Let's start at the beginning. What do you mean by a streaming loader?10:30
<ajs6f>awoods: A loader that takes in RDF triple-by-triple and puts it in the repo, maintaining an absolute minimum of state for speed and efficiency.10:31
awoods: Something you could reasonably pipe into on a Posix CL.10:32
<awoods>ajs6f: ok. Such a tool would expect that the URIs in the file have already been translated to F4 URIs?
<ajs6f>awood: Right.
awoods: There might be a tool for that, _but it's a separate tool._10:33
awoods; For one thing, it could accrue a _lot_ of state.
<awoods>called sed
<ajs6f>awoods: Works for me.
NTriples FTW!
<awoods>ajs6f: It would be great to use such a tool to test the capacity limits of F4.10:34
<ajs6f>awoods: I prefer to explore the incapacity limits.
<awoods>ajs6f: Hopefully that comes later.10:35
<ajs6f>awoods: Functionality is just a hangar for technical debt.
awoods: File a ticket for a streamer and I'll examine it.10:36
* awead joins
<awoods>ajs6f: will do.
<ajs6f>awoods: Seriously, I would do it in Scala. Half the code, all the speed, twice the type-safety.10:37
<awoods>ajs6f: why not10:38
<ajs6f>awoods: https://www.goodreads.com/quotes/37473-oh-well-mcwatt-sang-what-the-hell
awoods: Did you ever notice that Kendall Clark has Noam Chomsky as his Github avatar?10:39
<awoods>ajs6f: I do not know Kendall Clark10:40
<ajs6f>awoods: Half of what was Clark & Parsia (now Complexible). They did Pellet, the OWL reasoner, Stardog, a couple of other SemWeb things. He's contributed a lot to Banana RDF, W3C's Scala-RDF toolkit.10:41
awoods: I just think it's funny to imagine what Chomsky would think of that use of images.
<f4jenkins>Yippee, build fixed!10:52
Project fcrepo4-T2 build #277: FIXED in 4 min 36 sec: http://jenkins.fcrepo.org/job/fcrepo4-T2/277/
<ajs6f>awoods: Do you want this streaming parser VC'd into fcrepo-labs?12:02
<awoods>ajs6f: that would make sense
<ajs6f>awoods: Okay, I'm bumbling along. If I can get into the office I should have something more quickly.12:03
<awoods>ajs6f: obviously this is not urgent.12:04
<ajs6f>awoods: Nothing with Fedora is urgent.12:05
<awoods>ajs6f: new tagline?
<ajs6f>Fedora 4: What's _your_ hurry?
awoods: What's the current practice for getting a repo off-line? Use the repository.json directly?12:06
awoods: Use the SPring XML?
<awoods>ajs6f: migration-utils uses spring. https://github.com/fcrepo4-labs/migration-utils/blob/master/src/main/java/org/fcrepo/migration/Migrator.java12:09
<ajs6f>awoods: Okay, I'll do that. At least we can be consistent.
<awoods>ajs6f: same with fcrepo4-upgrade-utils: https://github.com/fcrepo4-labs/fcrepo4-upgrade-utils/blob/tech-md-migrator/src/main/java/org/fcrepo/upgrade/utils/TechnicalMetadataMigrator.java
<ajs6f>awoods: Wait, those don't help. Those are about separate Spring files to configure the utility. I'm talking about how to get access to the _repo_ config.12:10
<awoods>ajs6f: Both of these examples spin up the repository12:11
<ajs6f>awoods: Oh, wait, the second does that, from the classpath.
awoods: I'm not trying to spin up a repo— I'm trying to _not_ do that.
<awoods>ajs6f: you have what you need, no?12:12
<ajs6f>awoods: I'll figure it out.
awoods: We have no way to create a resource from an URI. You have to use an IdentifierConverter. So what's the assumption about bulk loading? Are people going to submit HTTP URLs?12:23
<awoods>ajs6f: I think HTTP URLs in the source for a bulk upload is a reasonable starting assumption.12:24
<ajs6f>awoods: Urg. That means I need to get from HTTP URIs to JCR Paths. And I can't just make an HttpResourceConverter because you need an UriInfo for that. maybe I can get an UriInfo from some obscure Jersey factory?12:28
Sorry, UriBuilder.
<awoods>ajs6f: Why build this into F4? and not just push against the REST API?12:29
<ajs6f>awoods: Because the whole point is to avoid HTTP. Or were you lying to me?
<awoods>ajs6f: the point was to not require users without scripting skills to have to parse out a bulk RDF document.12:30
<ajs6f>awoods: This is not an appropriate task for a REST endpoint. It specifically deals with bulk action on multiple resources. That is a terrible case for REST.12:31
<awoods>ajs6f: ok
whikloj: thanks for your work on the constraints issue... we are almost home.12:35
<whikloj>awoods: no problem, thanks for the patience as I kick it down the fairway.
<ajs6f>whikloj: It's your use of naked angle brackets in that file.12:36
whikloj: "roperty in the <http://fedora.info/definitions/v4/repository#> namespac"
whikloj: What's a fairway?
<whikloj>ajs6f: it's where other people play golf, and I drink and drive legally.12:37
<ajs6f>whikloj: Sounds like you get much the better end of the deal.
<whikloj>ajs6f: that might be up for debate, but I'm not good at debating either.12:38
awoods/ajs6f: PR updated with less '<'s12:42
<ajs6f>awoods: This bulk thing has got to take advatnage of the hierarchy-generating stuff, right?/12:46
<awoods>ajs6f: yes12:48
<ajs6f>awoods: And that's in a translation chain in the Spring XML?12:49
awoods I"m thinking I may just offer the option of chopping the "htttp://server/fedora/rest" prefix off the identifiers and taking what remains through the translation chain. That ought to do it, if the people using the script have used the right URLs.12:50
awoods: The advantage of doing it the right way is only this: it enforces the right URLs.
awood: My cheap way of doing it would allow people to import "http://wrongservername/fedora/stuff" without failing.12:51
<awoods>ajs6f: you are suggesting that imported resources would ignore the beginning part of the imported URLs? (http://host:port/domain)?12:52
<ajs6f>awoods: Basically.
<whikloj>ajs6f: why not check and then chop?12:53
<ajs6f>awoods: No, more like, if it matches a known-good value (configured) then ignore it. Otherwise use it.
<awoods>ajs6f: ok. How would the user specify the known value to expect/ignore?
ajs6f: tool configuration option?
<ajs6f>awoods: Yep12:54
<awoods>ok
<ajs6f>—repoPrefix=http://myrepo
or maybe —fedoraLocation=djgjdfg
* github-ff joins13:09
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/vtoHJ
fcrepo4/master 447e56a Jared Whiklo: POST/PATCH must advertise constraints if triples not persisted...
* github-ff leaves
* github-ff joins
[fcrepo4] awoods closed pull request #808: POST/PATCH must advertise constraints if triples not persisted (master...FCREPO-1418) http://git.io/vkHT8
* github-ff leaves
* ksclarke joins13:13
* awead leaves14:38
* awead joins
* awead_ joins15:40
* awead leaves15:42
* mhwood joins15:57
* dhlamb leaves16:07
* mhwood leaves17:10
* whikloj leaves17:54
* ksclarke leaves18:26
* github-ff joins19:03
[fcrepo-message-consumer] awoods opened pull request #85: Update test schema.xml due to impact of updated transform (master...fcrepo-1552) http://git.io/vtiZc
* github-ff leaves
* awead_away leaves19:30
* awead joins19:31
* ajs6f leaves19:51
* ajs6f joins19:54
* ajs6f leaves
* github-ff joins19:58
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/vtizB
fcrepo4/master aa8748b ajs6f: Java8 Refactoring...
* github-ff leaves
* github-ff joins
[fcrepo4] awoods deleted ALittleJava8 at b02cbd1: http://git.io/vtizu
* github-ff leaves
* github-ff joins20:13
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/vtiV2
fcrepo4/master 1ea8298 Andrew Woods: Merge pull request #821 from fcrepo4/ALittleMoreJava8...
* github-ff leaves
* github-ff joins20:26
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/vtiof
fcrepo4/master 1f29124 Andrew Woods: Merge pull request #823 from fcrepo4/FileConnectorPreJava8Stuff...
* github-ff leaves
* travis-ci joins20:34
fcrepo4/fcrepo4#3778 (master - 1ea8298 : Andrew Woods): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/aa8748ba908c...1ea829891b07
Build details : https://travis-ci.org/fcrepo4/fcrepo4/builds/68899554
* travis-ci leaves
* travis-ci joins20:35
fcrepo4/fcrepo4#3777 (master - aa8748b : ajs6f): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/447e56a95469...aa8748ba908c
Build details : https://travis-ci.org/fcrepo4/fcrepo4/builds/68898294
* travis-ci leaves
* travis-ci joins21:05
fcrepo4/fcrepo4#3779 (master - 1f29124 : Andrew Woods): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/1ea829891b07...1f2912474dad
Build details : https://travis-ci.org/fcrepo4/fcrepo4/builds/68900648
* travis-ci leaves
* dhlamb joins21:22
* awead leaves22:21
* awead joins22:32
* awead leaves22:40
* awead joins22:41
* dhlamb leaves22:44
* awead leaves23:06

Generated by Sualtam